Enhanced Visible-Light-Assisted Photocatalytic Removal of Tetracycline Using Co/La@g-C₃N₄ Ternary Nanocomposite and Underlying Reaction Mechanisms

Abstract

Graphitic carbon nitride (g-C₃N₄) is a promising material for photocatalytic applications. However, it suffers from poor visible-light absorption and a high recombination rate of photogenerated electron–hole pairs. Here, Co/La@g-C₃N₄ with enhanced photocatalytic activity was prepared by co-doping Co and La into g-C₃N₄ via a facile one-pot synthesis. Co/La@g-C₃N₄ displayed better performance, achieving 94% tetracycline (TC) removal within 40 min, as compared with g-C₃N₄ (BCN, 65%). It also demonstrated promising performance in degrading other pollutants, which was ~2–4-fold greater relative to BCN. The improved photocatalytic activity of Co/La@g-C₃N₄ was associated with improved photogenerated charge separation, reduced charge transfer resistance, a built-in electric field arising from the p-n-p heterojunction, and the synergistic effect of ternary components for the separation and transfer of the photogenerated charge carriers. Superoxide radicals are suggested to be the most notable reactive species responsible for the photocatalytic reaction. Environmental factors, including the pollutant concentration, catalyst dosage, solution pH, inorganic salts, water matrices, and mixture with dyes, were considered in the photocatalytic reactions. Co/La@g-C₃N₄ showed good reusability for five cycles of the photocatalytic degradation of TC. The facile one-pot co-doping of Co and La in g-C₃N₄ formed a p-n-p heterojunction with boosted photocatalytic activity for the highly efficient removal of TC from various water matrices.

1. Introduction

Since the discovery of Penicillin from Penicillium in 1928, there has been a growing number of antibiotics from synthetic and natural sources [1]. Antibiotics are broadly employed to treat or prevent diseases in humans and animals, and a proportion of them are used as growth promoters in animals [2]. Tetracycline (TC) is an antibiotic well known for its broad-spectrum activity against a wide range of Gram-positive and Gram-negative bacteria [3]. Despite the significant importance of TC, its residue in the environment or even in our water supply and irrigation systems has raised public concerns [3]. It may result in the growth of antibiotic-resistant bacteria species [3].

Due to the extensive application and poor adsorption of TC, most of it is excreted through faeces and urine as an unmetallized compound into wastewater. Meanwhile, many antibiotics, including TC, are not effectively removed by most wastewater treatment plants and, in turn, contaminate our ecosystem. Thus, there has been continuous research effort exploring effective techniques for efficient TC removal. Among the various approaches studied so far, photocatalysis has been demonstrated to be a promising and eco-friendly method that works by generating highly reactive radicals or oxidants to mineralize pollutants into harmless substances upon illumination [3].

Graphitic carbon nitride (g-C₃N₄) is one of the most widely studied photocatalyst materials owing to its stability, visible-light absorption, eco-friendliness, and cost-effectiveness [4,5,6]. Nevertheless, its use has been limited due to the high recombination rate of electron–hole pairs, the small surface area of the material, and the low absorption of visible light (<450 nm) [7]. Therefore, researchers have devoted efforts to modifying g-C₃N₄ using various strategies, including morphology tuning, elemental doping, and heterojunction fabrication [8]. Among them, metal doping into g-C₃N₄ has drawn significant research interest due to its ability to improve visible-light absorption, promote charge separation, reduce electron–hole recombination, and increase the active surface area, which resulted in enhanced performances of photocatalytic reaction [9]. Various metal dopants, including Ba, Er, and Eu, have been tested [10,11,12]. The resulting single-metal-modified g-C₃N₄ exhibited an enhanced visible-light response and reduced recombination of charge carriers relative to pure g-C₃N₄ for improved TC removal.

In recent years, the interest of researchers has moved to the co-doping strategy, which takes advantage of the potential synergistic effect of two dopants and, in turn, largely enables the tailoring of the photocatalytic activity of the base g-C₃N₄ catalyst [13]. So far, several combinations have been reported for dopant pairs consisting of a metal/non-metal or a non-metal/non-metal, e.g., B/Na, Bi/P, Ce/C, B/Eu, S/Bi, B/Se, S/P, and Mo/P [14,15,16,17,18,19,20,21], with some promising outcomes on the improved photocatalytic degradation of TC. However, the non-metal constituents of such dopant combinations could form recombination centres in the lattice of g-C₃N₄, thereby limiting the photocatalytic performance of the modified sample [22]. Some studies have also indicated the leaching of a non-metal dopant and thus the low stability of the modified photocatalyst [23]. On the other side, when adopting metal/metal co-doping, some metal atoms might replace C or N atoms in the g-C₃N₄ structure and, in turn, enhance the electron delocalization of the intralayer and change the band edge [24], whilst some might be incorporated into the interlayers of the g-C₃N₄ structures and then facilitate charge transfer [25]. Furthermore, the possibly formed unique heterojunction between dissimilar semiconductors can optimize absorption properties, enhance carrier separation, reduce electron recombination, accelerate charge transfer, and promote photocatalytic efficiency [26].

We have noted that, as compared to non-metal/non-metal and metal/non-metal co-doping, so far, the efforts contributed to developing and understanding metal/metal co-doping into g-C₃N₄ for the photocatalytic degradation of organics have been limited [25,27,28,29,30], with particularly little focus on pharmaceutical removal [31,32,33]. Among various metal dopants, the combination of rare-earth metals and transition metals has been explored for various photocatalytic applications [25,31,34,35,36,37]. A few have reported on the synthesis of visible-light-active photocatalysts targeting the elimination of pharmaceuticals and thus attracted our research interest. On the one side, cobalt (Co), as one of the popular transition-metal candidates for catalysts due to its high positive potential, has been proven to contribute to a lower energy band gap, a reduced recombination rate of electron–hole pairs, and the extended absorption of the solar spectrum [38,39]. On the other side, the rare-earth metal lanthanum (La) has shown a promising effect in interacting with functional groups of organics and enhancing the photocatalytic ability of g-C₃N₄ [40,41]. Co and La metals can augment the performance of g-C₃N₄ since they are able to capture photo-induced electrons and improve light absorption due to their unfilled 4f and empty 5d orbitals [12,42]. The co-doping of La and Co into g-C₃N₄ could boost the photocatalytic activity of the host material by providing a large number of active sites and lowering the activation energy of the photocatalytic system. However, the impact of La and Co dopants on the photocatalytic activity of bulk g-C₃N₄ for the visible-light-initiated photocatalytic removal of antibiotics has not yet been fully optimized, and the underlying mechanisms are not well understood.

Based on the above, in this study, we utilized a facile and cost-effective one-pot synthesis to prepare a Co₃O₄/g-C₃N₄/La₂O₃ ternary heterojunction composite (denoted by Co/La@g-C₃N₄). Visible light was used to initiate the photocatalytic reaction because of its broader range of the electromagnetic spectrum, which makes it more energy-efficient and cost-effective as compared with the utilization of UV light. It is also safer for the environment and human health when adopted for various photocatalytic applications. The visible-light-initiated photocatalytic activity of Co/La@g-C₃N₄ was evaluated for TC, MO, MB, and RhB removal. Afterwards, the photocatalytic degradation of TC was optimized by varying the pollutant concentration, solution pH, catalyst dosage, water matrices, and co-existing dyes and ions. The reusability of Co/La@g-C₃N₄ was investigated, along with an exploration of its photocatalytic mechanisms.

2. Materials and Methods

2.1. Synthesis and Characterization

Our photocatalyst samples were prepared via a facile one-pot synthesis. Typically, 10 g of urea ((NH₂)₂CO, 99.7% Sigma-Aldrich, Melbourne, Australia), 50 mg of lanthanum nitrate hexahydrate (La(NO₃)₃.6H₂O, 99.9%, Sigma-Aldrich, Melbourne, Australia), and 10 mg of cobalt hexachloride (CoCl₂.6H₂O, $\ge$97%, Sigma-Aldrich, Melbourne, Australia) were dissolved in 25 mL of ultra-pure water. The resulting suspension was sonicated for 30 min and then magnetically stirred at 80 °C for 1 h [43]. Afterwards, the solution was dried at 85 °C overnight in an oven, and then the obtained solid was calcined at 550 °C for 2 h (10 °C/min) in a furnace. The final product was designated Co/La@g-C₃N₄. For comparison, La@g-C₃N₄ and Co@g-C₃N₄ were prepared under the same reaction conditions as above but in the absence of CoCl₂.6H₂O and La(NO₃)₃.6H₂O, respectively. The sample of pristine g-C₃N₄ was synthesized by heating 10 g of (NH₂)₂CO at 550 °C for 2 h under ambient conditions [34].

X-ray diffraction (XRD) characterization at 2θ = 10–80° was performed by using a Rigaku SmartLab X-ray Diffractometer at a 10°/min scanning rate. The surface morphology was examined through a scanning electron microscope (SEM) using Tescan Clara (Brno, Czech Republic) and a transmission electron microscope (TEM) using TEM-TITAN (Boston, MA, USA). Fourier transform infrared resonance (FTIR, Perkin Elmer, Hopkinton, MA, USA) spectra of the synthesized samples were recorded between 4000 cm⁻¹ and 400 cm⁻¹ using a Perkin Elmer spectrophotometer. Nitrogen adsorption–desorption analysis was conducted at 77 K using Micromeritics Inc. SAPA 2010 (Norcross, Atlanta, GA, USA). The absorption spectra of the samples were measured using a Perkin Elmer Lambda 750 UV/Vis/NIR spectrophotometer, and the band gap was calculated from the Kubelka–Munk equation, (αhυ)² versus hυ.

2.2. Degradation Experiment

To evaluate the visible-light photocatalytic activity of the synthesized photocatalysts, an experiment on tetracycline (TC, 99.9%, Sigma-Aldrich, Melbourne, Australia) degradation was performed in a cylindrical Pyrex container, which was maintained at a constant room temperature under a Xenon lamp (CEL-HX F300, Beijing, China) with a 420 nm cut-off filter. Unless specified otherwise in the reaction conditions for the photocatalytic degradation of TC, the temperature was maintained at 25 °C, the initial TC concentration was 20 mg L⁻¹, the catalyst dosage was 0.4 g L⁻¹, and the initial pH of the TC solution was 5.05. In a typical experiment, 20 mg of the photocatalyst was dispersed into 50 mL of a TC aqueous solution. Before visible-light irradiation, the reaction solution was agitated for 30 min to reach adsorption equilibrium in the dark. Subsequently, the suspension was illuminated with the simulated visible light for 60 min, and 3 mL aliquots were taken at 10 min intervals. The collected solutions were centrifuged at 12,000 rpm for 5 min to separate photocatalyst solids from the solution. The residual TC concentration in the obtained solution was analysed using a Shimadzu UV/Vis spectrophotometer (Kyoto, Japan). Pollutant removal (%) was calculated using Equation (1), and the rate of degradation was obtained from the nonlinear pseudo-first-order kinetic model in Equation (2) [44,45]:

$$\mathrm{TC}\mathrm{removal}(\%)=\left(1-{\displaystyle \frac{C}{C_0}}\right)\times 100$$

(1)

$${\displaystyle \frac{-dc}{dt}}=kC$$

(2)

where C₀, C, k, and t denote the initial concentration (mg L⁻¹), the concentration at a given time (mg L⁻¹), the rate constant of degradation (min⁻¹), and the time of the reaction (min), respectively.

To investigate the applicability of our photocatalyst for TC removal, various dyes, including rhodamine B (RhB, ≥99.5%, Sigma-Aldrich, Melbourne, Australia), methyl orange (MO, 85%, Sigma-Aldrich, Melbourne, Australia), and methylene blue (MB, ≥81%, Sigma-Aldrich, Melbourne, Australia), were selected. Different water matrices were also collected, including tap water (TW), deionized water (DW), and rainwater (RW). Their characteristics are shown in Table S1. The TC solution was adjusted by adding either HNO₃ or NaOH to achieve a pH ranging from 2.0 to 11.5.

The stability and recyclability of the synthesized photocatalyst were analysed by reusing it for 5 consecutive cycles. After each run of photocatalytic degradation, the spent photocatalyst was separated from the solution, washed using water/ethanol, and, lastly, dried in an oven. The concentrations of leached metal ions were quantified using an inductively coupled plasma atomic emission spectrometer (ICP-AES) (Agilent 720 ICP-AES with SeaSpray nebulizer, glass-cyclonic Tracey spray chamber, Niagara Plus and inline filter (Glass Expansion); Temecula, CA, USA).

2.3. Photoelectrochemical Measurements

The photoelectrochemical measurements were performed using a Biologic SP-300 electrochemical workstation (Seyssinet-Pariset, France) with a typical three-electrode system. The reference, counter, and working electrodes were silver/silver chloride (Ag/AgCl), Pt wire, and the selected sample. The working electrode was fabricated as follows: 5 mg of the powder sample was mixed with 10 µL of Trixton-100 (Sigma Aldrich, Melbourne, Australia) and 20 µL of DI water to form a slurry. Subsequently, the slurry was coated on a tape-masked fluorine-doped oxide (FTO) glass slide using the doctor blade method to achieve an exposed active area of the sample of about 1 cm². The coated FTO slides were dried at 350 °C for 45 min. The transient photocurrent response was recorded in a 0.1 M Na₂SO₄ electrolyte solution in dark and light modes with an interval of 20 s under visible-light illumination (CEL-HX F300 with a 420 nm cut-off filter; Beijing, China). Electrochemical impedance spectroscopy (EIS) Nyquist plots were obtained over the frequency range of 50 mHz to 100 kHz. Analyses to obtain Mott–Schottky plots were executed at an AC voltage of 10 mV, a potential range of −1.0 V to +1.0 V, and frequencies of 100 Hz, 500 Hz, and 1000 Hz.

2.4. Identification of Active Radical Species and Possible Intermediates

To reveal the dominant active species, benzoquinone (BQ, >98%, Sigma-Aldrich), isopropyl alcohol (IPA, ≥99.0%, Sigma-Aldrich), AgNO₃ (≥99.9%, Sigma-Aldrich), sodium ethylenediaminetetraacetate (Na₂EDTA, ≥99.0%, Sigma-Aldrich), and L-Tryptophan (L-Trp., ≥98.0%) were introduced as scavengers for superoxide (•O₂), hydroxyl (•OH), electrons (e⁻), holes (h⁺), and singlet oxygen ¹O₂, respectively.

The possible intermediates generated were analysed using liquid column–mass spectrometry (LCMS) (Shimadzu LCMS-8045) with a C-18 capillary column (Shim-pack XR ODS III (1.6 µm), 2.0 id × 50 mm) (Kyoto, Japan) at 30 °C for the separation of TC intermediates. The mobile phase comprised HPLC-grade 70% water and 30% methanol with a flow rate of 0.4 mL and an injection volume of 0.5 µL. The mass spectra were obtained using the electrospray ionization (ESI) source at 15 kV. The masses of the intermediates were examined in the range of 100 m/z–500 m/z.

3. Results

3.1. Visible-Light Photocatalytic Activity of Photocatalysts

Figure 1a compares TC removal using Co/La@g-C₃N₄ via 30 min of adsorption in the dark followed by 40 min of photocatalytic degradation under visible-light irradiation to that using pristine g-C₃N₄, Co@g-C₃N₄, and La@g-C₃N_4. In the absence of a catalyst, the control photolysis experiment shows a negligible change in TC concentration, suggesting the good stability of TC molecules under our visible-light source. Pristine g-C₃N₄ exhibited a total TC removal of about 65% after photocatalytic degradation. The metal doping of g-C₃N₄ resulted in greater total TC removal via enhanced adsorption and photocatalytic degradation. In particular, the metallic co-doped photocatalyst (Co/La@g-C₃N₄) displayed a much better adsorptive capacity and photocatalytic activity through the synergy of Co and La dopants on g-C₃N₄ for fast and almost complete TC removal (94% within 30 min). The photocatalytic removal of TC closely follows pseudo-first-order kinetics (Figure 1b; R² ≥ 99.0%), and the rate constants (k) were calculated: Co/La@g-C₃N₄ (k = 0.28 min⁻¹) > La@g-C₃N₄ (k = 0.25 min⁻¹) > Co@g-C₃N₄ (k = 0.21 min⁻¹) > g-C₃N₄ (k = 0.19 min⁻¹).

Table 1. A comparison of photocatalytic performance for TC removal between Co/La@g-C₃N₄ and other metal-doped or co-doped g-C₃N₄ photocatalysts in the literature.

Photocatalysts (Synthetic Method)	Dosage (g/L)	TC (mg/L)	Light Source (Wavelength)	Removal (%) within Duration	Ref.
Cu-doped g-C3N4 (heating of pre-formed g-C3N4 and Cu)	0.2	50	1000 W Xe lamp (λ > 420 nm)	99 (30 min)	[46]
Ba-doped g-C3N4 (thermal condensation of Ba(NO3)2 and melamine)	0.5	10	150 W Xe lamp (λ > 400 nm)	70 (120 min)	[10]
Ag-doped g-C3N4 (photodeposition on pre-formed g-C3N4)	1.0	20	300 W Xe lamp (λ > 420 nm)	76 (120 min)	[47]
Eu-doped g-C3N4 (thermal treatment of pre-formed g-C3N4 and Eu2O3)	0.2	20	300 W Xe lamp (λ > 420 nm)	82 (50 min)	[12]
La-doped g-C3N4 (hydrothermal treatment of pre-formed g-C3N4 and La(NO3)3, followed by calcination)	-	10	Visible light (λ = 400–800 nm)	92 (90 min)	[40]
La2O3/g-C3N4 (mixing of pre-formed g-C3N4 and La2O3, followed by drying and heating)	2	20	Visible light (λ ≥ 400 nm)	40 (40 min)	[41]
B/Na co-doped g-C3N4 (mixing of pre-formed g-C3N4 and sodium borohydride, followed by calcination)	0.1	30	300 W Xe lamp (λ > 420 nm)	78 (30 min)	[14]
B/Eu co-doped g-C3N4 (thermal treatment of pre-formed g-C3N4, H3BO3 and Eu2O3)	0.4	20	400 W Halogen lamp	91 (50 min)	[48]
S/Se co-doped g-C3N4 (mixing of pre-formed g-C3N4 and SeS2, followed by heating)	0.4	30	500 W Xe lamp (λ > 420 nm)	78 (60 min)	[19]
Co/La@g-C3N4 (one-pot synthesis)	0.4	20	300 W Xe lamp (λ > 420 nm)	94 (40 min)	This study
	0.4	50	300 W Xe lamp (λ > 420 nm)	88 (40 min)

shows the competitive photocatalytic TC degradation performance of Co-Co/La@g-C₃N₄ under two TC concentration regimes in comparison with other metal-doped or co-doped g-C₃N₄ photocatalysts reported in the literature. It should be noted that different light sources and reaction conditions (e.g., catalyst dosage, TC concentration, solution pH, and reaction time) were utilized in the literature. However, it is still obvious that our Co/La@g-C₃N₄ shows good photocatalytic activity for fast and high TC removal from water, taking into consideration the selected light source, solution pH, catalyst dosage, TC concentration, and reaction time.

To explore the extended use and universality of the photocatalytic removal of organics by Co/La@g-C₃N₄, three dye targets, i.e., MO, MB, and RhB, as representative anionic, cationic, and zwitterionic dyes, were also tested, as shown in Figure 1c. Significant amounts of the organics were degraded within 40 min of visible-light irradiation, with the highest adsorptive removal in MB solution and the highest rate of degradation in RhB solution. Pristine g-C₃N₄ shows a removal percentage of 35%, 27%, and 62% for MO, MB, and RhB, respectively. As compared with pristine g-C₃N₄, the total removal % was improved to 75%, 86%, and 79% for MO, MB, and RhB, respectively, attributed to considerably improved adsorption and enhanced photocatalytic degradation (Figure 1c,d). The above results verify that Co/La co-doping into g-C₃N₄ can effectively improve the visible-light photocatalytic activity of g-C₃N₄, and the synthesized Co/La@g-C₃N₄ could potentially be an efficient photocatalyst to treat wastewater containing different organics under visible-light illumination. Moreover, the facile one-pot synthesis of our photocatalyst with good treatment capacity for the removal of various dyes would be advantageous for its practical application in wastewater treatment. To thoroughly explore its applicability and understand the reaction mechanisms, the characterization of the material and its operation optimization were carried out.

3.2. Mechanism of Enhanced Visible-Light Photocatalytic Activity

3.2.1. Material Property Characterization

In order to better understand the mechanisms governing the enhanced visible-light photocatalytic activity of Co/La@g-C₃N₄, variations in the physicochemical and spectroscopic properties of the synthesized photocatalyst were examined in detail and compared with those of g-C₃N₄, La@g-C₃N₄, and Co@g-C₃N₄.

The FTIR spectra of g-C₃N₄, La@g-C₃N₄, Co@g-C₃N₄, and Co/La@g-C₃N₄ are compared in Figure 2a. All of the photocatalyst samples display similar vibrational patterns [49]. The characteristic band found at 810 cm⁻¹ is assigned to the breathing mode of the main structural moiety of g-C₃N₄—the triazine ring. The stretching vibrations of g-C₃N₄ C–N heterocycles are marked by peaks in the range 1210–1650 cm⁻¹. The broad peaks at around 3200 cm⁻¹ are attributed to N-H stretching vibration. No noticeable differences are seen in Figure 2a for all photocatalysts; this suggests that the mono-doping or co-doping of g-C₃N₄ herein did not alter the chemical structure of g-C₃N₄, possibly due to the small amounts of dopants introduced.

In Figure 2b, an overall reduction in the intensity of characteristic XRD peaks corresponding to the (100) and (002) crystal planes of g-C₃N₄ is observed for all doped g-C₃N_4. Specifically, the peak at 13.1°, indexed to the (100) facet and ascribed to the in-plane repeating tris-s-triazine units, diminishes upon doping [50]. Smaller crystallite sizes, calculated using the Scherrer equation, are shown in

Table 2. Crystalline sizes and textural properties (pore diameter, pore volume, and BET specific surface area) of g-C₃N₄, La@g-C₃N₄, Co@g-C₃N₄, and Co/La@g-C₃N₄.

Photocatalyst	Crystalline Size (nm)	Pore Diameter (nm)	Pore Volume (cm3 g−1)	Specific Surface Area (m2 g−1)
g-C3N4	4.77	18.6	0.0124	41.3
La@g-C3N4	2.02	20.5	0.0135	45.8
Co@g-C3N4	2.57	20.9	0.0241	79.8
Co/La@g-C3N4	2.08	19.2	0.0192	65.5

for all doped g-C₃N₄ variants_. These suggest that the introduction of metal dopants interferes with the thermal polymerization of precursors forming g-C₃N₄ and then leads to a reduction in the ordered structure in the framework [51]. There are no apparent new peaks associated with Co or La species in any of the doped g-C₃N₄ variants, which might be explained by the small amounts of dopants added.

The textural properties, including porosity and BET specific surface area, of various samples are detailed and summarized in Table 2; their nitrogen adsorption–desorption isotherm curves are included in Figure 2c. All of the samples display typical IV isotherm curves and H3-hysteresis loops, suggesting the presence of silt-like mesopores [52]. Overall, Co- or/and La-doped g-C₃N₄ exhibited increases in pore volume, pore diameter, and BET specific surface area with respect to g-C₃N₄, which could be due to the altered polymerization of urea in the presence of different metal dopants. Such features might be beneficial in promoting the transport of pollutants into photocatalyst pores and then their arrival at active sites for reactions. The sample of Co@g-C₃N₄ displayed the most significant increase in BET specific surface area (79.8 m² g⁻¹) as compared to g-C₃N₄ (41.3 m² g⁻¹). The increase in the surface area of Co@g-C₃N₄ may be ascribed to the creation of porous structures induced by co-doping during polymerization [53]. This is also reflected in the increase in the pore diameter and pore volume (20.9 nm and 0.0241 cm³ g⁻¹ for Co@g-C₃N₄ vs. 18.6 nm and 0.0124 cm³ g⁻¹ for g-C₃N₄). The surface area and pore volume of La@g-C₃N₄ are comparable to those of g-C₃N_4, but they show a slightly larger average pore diameter. The reduced pore diameter of the co-doped photocatalyst (Co/La@g-C₃N₄) compared to the mono-doped samples (La@g-C₃N₄ or Co@g-C₃N₄) could be due to the pore and channel obstruction by the dopants [54].

The morphology of g-C₃N₄, as shown in Figure 3a, is characterized by the stacked non-uniform sheet-like morphology of g-C₃N₄, which ensued from the thermal polymerization of urea. Upon doping with Co and La, a graphitic-like morphology with non-uniform sheets and obvious porosity is observed in the modified sample, which are believed to be advantageous features for the photocatalytic degradation of organic pollutants.

Figure 4 details the microstructure of Co/La@g-C₃N_4, with a typical lamellar structure (Figure 4a) and the distribution of dopant particles on it (Figure 4b). The HRTEM images (Figure 4c,d) show lattice fringes with a spacing of 0.15 nm and 0.29 nm, corresponding to the (511) and (011) planes of Co₃O₄ and La₂O₃, respectively [55,56]. The high-angle annular dark-field (HAADF) image (Figure 4e) confirms the presence of C, N, O, Co, and La in Co/La@g-C₃N₄; the distribution of each constituent element is presented in Figure 4e-i–e-v. It is noteworthy that the detected O atom was incorporated during the thermal polymerization of urea [57]. TEM and SEM observations indicate that upon the introduction of Co and La, Co₃O₄ and La₂O₃ grow on the surface of g-C₃N₄, forming a p-n-p heterojunction and later showing improved photocatalytic activity for pollutant removal.

The XPS survey spectrum (Figure 5a) shows signals of N, O, and C, along with Co and La. The deconvoluted C 1s spectrum shows three main peaks at 284.5 eV, 287.8 eV, and 290.3 eV, which are attributed to adventitious C-C/C=C, -C-N/-C=N, and -O-C=N functional groups, respectively (Figure 5b) [58,59]. The deconvolution of the N 1s spectrum resulted in three peaks at binding energies of 399.1 eV, 401.4 eV, and 403.7 eV, which are ascribed to the nitrogen atoms of C=N-C, N-(C)₃, and C-NH₂, respectively (Figure 5c) [60]. The XPS spectra of O 1s with two peaks at binding energies of 532.5 eV and 536.6 eV suggest oxygen atoms in C-O and loosely bonded oxygen species on the surface, e.g., adsorbed water molecules (Figure 5d) [61]. The XPS spectrum of Co 2p can be deconvoluted into Co 2p_3/2 and Co 2p_1/2 with binding energies (satellite peaks) of 779.5 eV (sat. 786.9 eV) and 794.1 eV (sat. 803.7 eV), respectively (Figure 5e), suggesting the presence of Co²⁺ and Co³⁺ [62,63]. The XPS spectrum of La 3d revealed two peaks at 840.0 eV and 857.1 eV, which are attributed to La 3d_5/2 and La 3d_3/2, respectively (Figure 5f). The spin–orbit coupling between these two peaks was calculated to be 17.1 eV, which is similar to other reported work [55]. Our XPS results further support the successful co-doping of La and Co into g-C₃N₄ in the sample of Co/La@g-C₃N_4, which was observed in the above TEM and EDX images.

Figure 6a reveals an increase in visible-light absorption in the samples of La@g-C₃N₄, Co@g-C₃N₄, and Co/La@g-C₃N₄ as compared to the parent g-C₃N₄, with an absorption band edge around 465 nm. The samples of Co@g-C₃N₄ and Co/La@g-C₃N₄ exhibited a further shift in absorption to about 600 nm. The extra absorption shoulder, especially exhibited by Co/La@g-C₃N_4, in the far-visible region might be ascribed to n–π* transitions in the conjugated system of g-C₃N₄ when the structural distortion was developed [64]. The improved absorption in the visible-light range might also be due to the synergistic effect of the two dopants and g-C₃N_4. The band gaps (E_g) of the parent g-C₃N₄, La@g-C₃N₄, Co@g-C₃N₄, and Co/La@g-C₃N₄ were measured by adopting the Kubelka–Munk equation and are shown in Figure 6b, with values of 2.74 eV, 2.60 eV, 2.50 eV, and 2.62 eV, respectively.

The photocurrent response and electrochemical impedance spectroscopy (EIS) data were collected to investigate the separation and transfer of photogenerated charge carriers over our as-prepared photocatalysts. Figure 6c shows the order of the photocurrent density: g-C₃N₄ < Co@g-C₃N₄ < La@g-C₃N₄ < Co/La@g-C₃N₄. In particular, the photocurrent density of Co/La@g-C₃N₄ increases rapidly and is much higher than that of the parent g-C₃N₄, indicating that the synergistic effect of co-doped Co and La on g-C₃N₄ results in the fast separation of photogenerated charge carriers. The EIS Nyquist plots (in Figure 6d) suggest the charge transfer resistance of the photocatalysts; generally, a smaller arc radius represents a lower interfacial transfer resistance and greater electron mobility. As compared with g-C₃N₄, the interface resistance of Co/La@g-C₃N₄ is reduced by the synergistic interaction between dopants and g-C₃N₄, which promotes charge separation and, in turn, enhances photocatalytic degradation performance.

The pristine g-C₃N₄ and Co/La@g-C₃N₄ band-edge potentials were determined from the Mott–Schottky analysis at different frequencies (100 Hz, 500 Hz, and 1000 Hz). The positive slopes in the Mott–Schottky plots of both samples suggest n-type semiconductor features [65]. By extrapolating 1/C² = 0 in the plots (Figure 6e,d), the flat-band potentials of g-C₃N₄ and Co/La@g-C₃N₄ were deduced to be −0.64 eV and −0.50 eV (vs. Ag/AgCl electrode), respectively. In general, the flat-band potential is approximately equal to the conduction band potential (E_CB), and it is 0.1 eV more negative than the band potential for n-type semiconductors [66,67]. Thus, the E_CB values of g-C₃N₄ and Co/La@g-C₃N₄ are −0.74 eV and −0.60 eV, equivalent to −0.54 eV and −0.40 eV versus the normal hydrogen electrode (i.e., E_NHE = Ag/AgCl + 0.197). The valence band potentials (E_VB) of g-C₃N₄ and Co/La@g-C₃N₄ were calculated to be +2.20 eV and +2.22 eV, respectively, using the equation E_VB = E_CB + E_g [21]. The following results indicate a band gap narrowing with a positive shift in E_CB and E_VB edges of g-C₃N₄ upon co-doping with Co and La.

3.2.2. Routes of Active Species Formation and Possible Pathways of Degradation

To identify active species involved in the photocatalytic degradation of TC over Co/La@g-C₃N₄, reagents, including BQ, IPA, Na₂EDTA, L-Trp., and AgNO₃, were utilized to scavenge superoxide radicals (•O₂⁻), hydroxyl radicals (•OH), holes (h⁺), singlet oxygen (¹O₂), and electrons (e⁻), respectively. Figure 7a compares the removal of TC in the presence of different scavenging agents against the control in the absence of any agent. When IPA is introduced as a capture agent, the removal of TC is slightly impeded, indicating a minor role of •OH during the degradation reaction. The TC removal percentage is reduced to about 53% when Na₂EDTA is added, showing an important role of h⁺ in the overall photocatalytic degradation. The lowest TC removal % (13%) is noticed upon adding BQ, suggesting •O₂⁻ as the main ROS responsible for the removal of TC over Co/La@g-C₃N₄. The presence of ¹O₂ is deduced via the retarding of the TC removal effect of the L-Trp. reagent. The coexistence of ¹O₂ in the photocatalytic reaction system may be associated with the h⁺-induced oxidation of •O₂⁻ [68]. The synergy in generating electrons by the co-doped La and Co atoms was examined by introducing AgNO₃ as a capture agent. Only about 1% of TC was degraded when the photogenerated e⁻ was captured by AgNO₃, showing the important role of e⁻ in the generation of the main ROS (•O₂⁻).

Based on the above scavenging experiment, the TEM microstructure, and electronic band potential edges of Co/La@g-C₃N_4, a possible degradation mechanism is proposed (Figure 7c). The as-prepared photocatalyst was shown earlier to consist of Co₃O₄ and La₂O₃ on the surface of g-C₃N₄. g-C₃N₄ is an n-type semiconductor, whilst both Co₃O₄ and La₂O₃ are p-type semiconductors [63,69]. With the newly constructed Co/La@g-C₃N₄ ternary composite photocatalyst, a p-n-p heterojunction is formed among g-C₃N₄ and the two types of metal oxides.

For a concise understanding of the heterojunction formation with g-C₃N₄, the electronic band structures of the metal oxides were investigated using UV-Vis DRS and Mott–Schottky analysis. The band gap (Figures S2 and S3) and conduction band (CB) from the Mott–Schottky plots (Figures S4 and S5) of Co₃O₄ and La₂O₃ were measured to be 2.03 eV (VB = +0.99 V (+1.18 eV vs NHE)) and 3.68 eV (VB = +1.90 V (+2.09 eV vs NHE)), respectively. The corresponding valence bands (CBs) of Co₃O₄ and La₂O₃ were separately computed to be -0.85 eV and -1.59 eV. The band-edge potentials of the respective semiconductors before contact are shown in Figure 7b. The Fermi level (E_F) of Co₃O₄ or La₂O₃ is located just above the VB top for p-type semiconducting materials, whilst that of g-C₃N₄ is situated just below the CB bottom. In the p-n-p junction formation and Fermi level equilibration, holes are migrated from the VB of the p-type semiconductors (Co₃O₄ and La₂O₃) to CB of the n-type semiconductor (g-C₃N₄) because of concentration differences [67]. At the same time, electrons diffuse from the CB of g-C₃N₄ to the CB of (Co₃O₄ and La₂O₃), giving rise a positively charged g-C₃N₄ and negatively charged (Co₃O₄ and La₂O₃). As reported in the literature, an internal electric field forms between g-C₃N₄ and metal oxides in the Fermi level equilibration of p-type and n-type semiconductors, resulting in a shift in their band positions [70]. As shown in Figure 7c, the energy bands of metal oxides shift upward along the Fermi levels; the energy bands of g-C₃N₄ shift downward along the Fermi levels. Following visible-light irradiation, photogenerated electrons are excited to the respective CBs of the two visible-light-responsive semiconductors (g-C₃N₄ and Co₃O₄), leaving holes in the VBs (Figure 7c). Electrons are transferred from the CB of p-type Co₃O₄ to n-type g-C₃N₄ and holes migrate in the reverse direction. Additionally, the co-doping of Co and La aided the separation of the photogenerated charge carriers and minimized the charge transfer resistance of the photocatalyst, as observed in Figure 6c,d. On the other side, La₂O₃, as a visible-light-inactive metal oxide, cannot be excited under visible-light illumination due to its large band gap. At the junction interface of g-C₃N₄ and Co₃O₄, electrons are diffused from the CB of Co₃O_4, which is more negative, to the CB of g-C₃N₄, leaving holes in the VB for the direct oxidation of TC molecules into their smaller byproducts. Consequently, the accumulated electrons in the CB of g-C₃N₄ are transferred to the surface of a visible-light-inactive La₂O_3, where they react with dissolved oxygen in the aqueous solution to generate •O₂⁻ (as evidenced in Figure 7a). Most organic pollutants are not degraded directly by •O₂⁻ due to its low oxidation potential. Thus, it is transformed into stronger oxidants like •OH and ¹O₂ by reacting with water and holes, respectively, which are utilized for TC degradation [71]. The synergistic effect among g-C₃N_4, Co₃O₄, and La₂O₃ facilitates electron separation and transfer in the ternary nanocomposite and, in turn, results in the accumulation of electrons on the photocatalyst surface, as deduced from the scavenging test when using AgNO₃ as an agent to capture electrons (Figure 7a). The tertiary Co₂O₃/g-C₃N₄/La₂O₃ in Co/La@g-C₃N₄ plays a vital role in boosting the photocatalytic activity of the catalyst.

Plausible pathways are proposed for TC degradation using the as-prepared Co/La@g-C₃N₄, as shown in Figure 8. The TC molecule (in Figure 8) structurally consists of various ionizable functional and high-electron-density groups, such as phenol, amino, and conjugated double bonds. It is prone to be attacked by reactive species—^•O₂⁻, ¹O_2, ^•OH, or h⁺, which were proven to be present in the scavenging test (Figure 7a)—to generate the identified intermediates—peaks P1–P11 with m/z consistent with the literature [72,73,74,75], as shown in the LC–MS analysis (Figure S6). It is noteworthy that since the main ROS, ^•O₂⁻, has a lower oxidation potential than most organic pollutants, it may be converted to ^•OH or ¹O₂ species with a stronger oxidation potential for degrading TC molecules [71]. Based on the literature [71,72,73,74,75,76,77,78] and the identified main ROS (^•O₂⁻), we propose two degradation routes, as illustrated in Figure 8. In pathway I, a primary intermediate (P1) with m/z = 476.45 can be generated through hydroxylation at C-11a and C-2 of the TC molecule (m/z = 445.25) [76]. Subsequently, P1 may be fragmented into P2 (m/z = 453.15) via dehydroxylation, and a further loss of the hydroxyl group at C-6 could produce P3 (m/z = 425.20). The peak P4 (m/z = 339.05) may be obtained from P3 via demethylation, dehydroxylation, and deamination [74]. The subsequent demethylation of P4 could transform it into P5 (m/z = 308.30) [77]. In pathway II, the TC molecule could be directly attacked to produce P7 (m/z = 400.00) via a deacylation reaction at C-1 and C2 of the TC molecule and further fragment to give P6 (m/z = 263.05) [72]. The intermediates P5 and P6, due to continuous attack by the ROS, could fragment into a series of lower-molecular-weight intermediates, including P8 (m/z = 165.95), P9 (m/z = 149.95), P10 (m/z = 128.10), P11 (m/z = 105.05), CO₂, and water molecules, through ring opening and molecular rearrangement [78].

3.3. Effects of Reaction Conditions on Photocatalytic Degradation

Figure 9a shows the effect of the catalyst dosage (0.1 g L⁻¹–0.6 g L⁻¹) on TC removal using Co/La@g-C₃N_4. TC removal increases from 80% at a dosage of 0.1 g L⁻¹ to 88% at a dosage of 0.4 g L⁻¹. The improvement in photocatalytic performance, which is evident from the higher rate constants with greater sample loading, may be due to the increasing number of exposed active sites on the photocatalyst for the adsorption of TC and the absorption of light, as well as the generation of reactive species for photocatalytic TC degradation [79]. On the other side, excessive loading (e.g., at a dosage of 0.6 g L⁻¹) results in a decline in the removal rate of TC (80%), which is believed to be associated with greater turbidity of the solution, preventing light penetration through aqueous media [80]. Furthermore, a high catalyst dosage may lead to the agglomeration of photocatalyst particles, the reduced availability of active sites on the catalyst surface, and, eventually, a decrease in the number of reactive species for the photocatalytic reaction [79,80]. As can be seen from our results, a dosage of 0.4 g L⁻¹ was suggested as the optimal dose for photocatalytic TC degradation over Co/La@g-C₃N₄ in subsequent experiments.

The initial TC concentration was varied from 20 mg L⁻¹ to 100 mg L⁻¹ to study the impact of pollutant concentration on its removal (Figure 9b). Prior to light illumination, the adsorption in the dark decreases with increasing TC concentration, which may be due to the excessive number of adsorbates on catalyst adsorptive sites [81]. Upon illumination, the photocatalytic removal of TC over Co/La@g-C₃N₄ varies with the initial TC concentration, especially within the first 20 min of the reaction. The value of k is higher at lower concentrations of TC, e.g., 20 mg L⁻¹ TC (0.28 min⁻¹), as compared with 50 mg L⁻¹ TC (0.15 min⁻¹), 80 mg L⁻¹ TC (0.10 min⁻¹), and 100 mg L⁻¹ TC (0.05 min⁻¹). This decline in the photocatalytic performance of Co/La@g-C₃N₄ may be due to the occupation of active sites by an excessive amount of TC (or produced intermediate products) and then the reduced formation of reactive species and photocatalysis reaction [82,83]. Moreover, the absorption of light by a greater amount of TC or its intermediates in the solution would hinder photon absorption on the catalyst active sites and, in turn, might deteriorate the photocatalytic activity of Co/La@g-C₃N₄ [79]. As can be seen, the increase in the initial TC concentration from 20 mg L⁻¹ to 100 mg L⁻¹ results in a corresponding decrease in TC removal from 88% to 73%, which aligns with the literature [79].

The solution pH is an important factor for optimizing photocatalytic degradation, as it affects the surface charge of the catalyst and the ionic form of the pollutant [84]. Figure 9c shows the removal of TC using Co/La@g-C₃N₄ at different pHs. The highest adsorption in the dark is recorded at pH = 5.05, whilst the highest rate constant (k) is found at pH = 11.23 (k = 0.35 min⁻¹). An optimal total TC removal of 88% is observed at pH = 5.05. A further increase in pH resulted in a corresponding decrease in the removal of TC from 76% (pH = 9.06) to 66% (pH = 11.23). The point of zero charge (PZC) of Co/La@g-C₃N₄ is shown in Figure S1, 6.6. This implies that at pH > pH_PZC, the catalyst surface is negatively charged, whilst at pH < pH_PZC, it gives rise to a positively charged surface. It is known that TC molecules exist in dissimilar forms depending upon the pH of the solution [85], including cationic species (pH < 3.3), zwitterionic species (3.3 < pH < 7.68), or anionic species (pH > 7.68). In our study herein, as compared with lower pH, the faster degradation rate at higher pH might be ascribed to the unstable nature of TC in alkaline conditions [86]. Meanwhile, at higher pH, the electrostatic repulsion between the negatively charged surface of the catalyst and the anionic form of TC molecules, as well as the competition between hydroxide ions and pollutants for active sites of the catalyst, might both contribute to the observed lower adsorption in the dark and then overall TC removal [86]. On the other side, acidic conditions (e.g., pH = 2.02) are not favourable for the adsorption of the cationic form of TC molecules onto the positively charged catalyst surface due to electrostatic repulsion [87]. Thus, lower adsorptive removal and the subsequent photocatalytic degradation of TC are seen when using Co/La@g-C₃N_4.

Organic pollutants are usually found to exist with many foreign ions in wastewater or water. Figure 10a shows the specific removal of TC using Co/La@g-C₃N₄ in the presence of some of the most common ions (e.g., Cl⁻, SO₄²⁻, or NO₃⁻). As compared with the control (in the absence of foreign ions), the adsorptive and photocatalytic removal of Co/La@g-C₃N₄ is noticed to vary insignificantly (~5% variation). The order of k change is NO₃⁻ (k = 0.22 min⁻¹) > Cl⁻ (k = 0.17 min⁻¹) > SO₄²⁻ (k = 0.16 min⁻¹), relative to the control (k = 0.15 min⁻¹). A plausible explanation for this phenomenon may be attributed to the possible formation of the radicals Cl•, NO₃•, and SO₄⁻ in the presence of the corresponding inorganic ions [84,88]. Our results demonstrate the robustness of Co/La@g-C₃N₄ for the efficient removal of TC in the presence of different inorganic ions.

Figure 10b compares TC removal using Co/La@g-C₃N₄ from different water matrices: deionized water (DW), rainwater (RW), and tap water (TW)_. As compared with TW and DW, lower adsorption of TC in the dark but greater photocatalytic removal of TC is seen in RW. The total TC removal changed insignificantly, which was 89% in TW, 88% in DW, and 85% in RW. The rate constant of degradation (k) is in the increasing order TW (0.09 min⁻¹) < RW (0.11 min⁻¹) < DW (0.15 min⁻¹). Figure S7 shows the total removal of TC from the primary (PWW) and secondary effluents (SWW), which were collected from a Perth local municipal wastewater treatment plant, reaching ~80% and 75% after 30 min adsorption and 60 min visible-light photocatalytic degradation, respectively. In general, our prepared Co/La@g-C₃N₄ exhibited good photodegradation activity in various water matrices.

Organic pollutants always exist along with other pollutants in real wastewater. Figure 10c investigates the versatility of Co/La@g-C₃N₄ in a binary solution consisting of TC and different dyes (e.g., MO, MB, and RhB). The total TC removal using Co/La@g-C₃N₄ decreases to about 73%, 84%, and 82% in binary mixtures of TC/MO, TC/MB, and TC/RhB, respectively, as compared to the TC-only solution (88%). The values of k for photocatalytic TC degradation in the systems of TC/MO, TC/RhB, and TC/MB are 0.05 min⁻¹, 0.18 min⁻¹, and 0.30 min⁻¹, respectively, in comparison with 0.15 min⁻¹ in the TC-only system. The reduced removal of TC and dyes from the binary systems, in comparison with the dye-only solution systems, could be due to competition for the adsorptive sites of Co/La@g-C₃N₄ [89]. Additionally, the effect of binary systems on the removal of dye/TC mixtures as compared to the dye-only solutions was investigated (Figure 10d). The significant removal performance of all dye-only solutions is noticed after 60 min of visible-light illumination. However, the degradation % of the respective dye is impeded in a binary system, which could be due to the electrostatic interactions between various ionic species of TC and the dye molecules. In the dark reaction, a decrease in adsorption can be observed in dye/TC binary systems relative to the dye-only solution, which affects the overall removal performance of the dye molecules. The rate of TC removal can also be affected by hydrogen bonding between the dyes and TC molecules and their electrostatic interactions with the surface of Co/La@g-C₃N₄ [90].

Relatively good reusability of Co/La@g-C₃N₄ can be achieved with five cycles of photocatalytic TC degradation, during which the degradation rate is reduced by less than 10%, as shown in Figure 11a. The stability of Co/La@g-C₃N₄ in the reaction is supported by the almost unchanged XRD patterns and FTIR spectra of fresh Co/La@g-C₃N₄ and spent Co/La@g-C₃N₄ (after fifth cycle of use), respectively. The ICP-AES analysis shows negligible amounts of Co (0.84 mg/kg) and La (2.8 mg/kg) in the solution after the photocatalytic reaction, which suggests that approximately 0.02% and 0.01% of the actual amount of Co (4500 mg/kg) and La (28,000 mg/kg) dopants leach out into the solution from the photocatalyst. This further proves the stability of Co/La@g-C₃N₄ for recycling and reuse.

4. Conclusions

A facile one-pot pyrolysis method was utilized by mixing urea and Co- and La-containing precursors to synthesize a g-C₃N₄-based ternary photocatalyst (Co/La@g-C₃N₄) with a unique p-n-p heterojunction. Upon visible-light illumination, Co/La@g-C₃N₄ displayed improved performance, achieving 94% TC removal, as compared with 65% of the pristine g-C₃N₄. The enhanced photocatalytic activity of Co/La@g-C₃N₄ was ascribed to the improved photogenerated charge separation, reduced charge transfer resistance, built-in electric energy arising from the p-n-p heterojunction, and synergistic effect of ternary components for the generation and transfer of electrons. Additionally, Co/La@g-C₃N₄ showed excellent performance under varying reaction conditions, including the solution pH, catalyst dosage, TC concentration, co-existing ion, and water matrix. Photocatalytic TC degradation by Co/La@g-C₃N₄ could be largely maintained after the fifth cycle of the reaction, confirming its good stability and reusability. The capture experiment demonstrated that •O₂⁻ played the most important role in the photocatalytic reaction. The facile fabrication and high performance of the Co/La@g-C₃N₄ ternary nanocomposite in this work suggest the promising potential for its practical application as a highly efficient visible-light-active photocatalyst.